Graphical abstract graphic file with name fx1.jpg [43]Open in a new tab Highlights * • 3D spheroid cultures were established in 384-well format * • Eight different media variants were used to optimize the 3D cultures * • Optimized William's medium was as good as expensive commercial medium * • The 3D cultures were used to study drug metabolism and toxicity __________________________________________________________________ Biological sciences; Proteomics; Methodology in biological sciences Introduction Hepatocytes comprise almost 80% of the liver volume, where they perform important functions including drug metabolism and xenobiotic detoxification ([44]Kmieć, 2001; [45]Trefts et al., 2017). Isolated primary human hepatocytes (PHH) are therefore considered the gold standard for studying hepatic metabolism and toxicity in vitro ([46]Gómez-Lechón et al., 2014; [47]Weaver et al., 2020). However, PHH are highly dependent on a finely-tuned in vivo microenvironment, which is difficult to reproduce in vitro. PHH rapidly loses their liver-specific functions in conventional 2D cultures, limiting their application to short-term studies ([48]Bell et al., 2018). One way to maintain long-term function of PHH is to culture grow them in 3D format ([49]Weaver et al., 2020; [50]Zhou et al., 2019). When 3D cultures are used, the key phenotypic traits can be maintained for a longer time time in culture. 3D cultures range from rather simple self-assemblies of PHH to more intricate co-culture systems such as scaffold-based and micro-patterned ([51]Weaver et al., 2020; [52]Zhou et al., 2019). Among the various 3D culture formats, PHH spheroids have been particularly popular in studies of hepatic drug metabolism and toxicity ([53]Bell et al., 2016; [54]Li et al., 2020; [55]Mizoi et al., 2020). Establishment of PHH spheroid cultures is fairly straightforward. In theory, their technical reproducibility should be good given that the individual spheroids are evenly sized and formed from the same number of hepatocytes ([56]Messner et al., 2013). However, the reproducibility of hepatocyte function varies greatly across laboratories. For instance the metabolite formation rates vary greatly ([57]Arakawa et al., 2017; [58]Bell et al., 2018; [59]Foster et al., 2019). A reason could be that different culture conditions are used. Cell culture media for PHH spheroids range from well-defined media like William's E ([60]Bell et al., 2016; [61]Rowe et al., 2013; [62]Xiang et al., 2019), Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F12) ([63]Dong et al., 2016), and Advanced DMEM/F12 ([64]Hu et al., 2018) to commercial ones such as Hepatocyte Maintenance Medium (HMM) with undisclosed contents ([65]Buesch et al., 2018; [66]Forsythe et al., 2018; [67]Ölander et al., 2019). The media also vary in glucose and insulin levels, ion composition, and nutrient content. We therefore set out to investigate how such media differences influence the maintenance of PHH phenotype and liver-specific functions in 3D long-term spheroid cultures with the aim to select optimal culture conditions for drug metabolism and toxicity studies. After establishing conditions for spheroid cultures in a 384-well format, we followed spheroid formation and phenotype using immunohistochemistry, ATP content and albumin production for three weeks. Signaling pathways and epithelial-mesenchymal transition (EMT) were investigated using quantitative global proteomics. Finally, we focused on ADME proteins relevant for drug metabolism and toxicity in functional studies. The results indicated that optimal PHH function occurs within two weeks of culture, physiological levels of glucose and insulin are preferable, and different media have a substantial impact on the culture. Results Optimizing spheroid formation in a 384-well format To facilitate screening of drug metabolism and toxicity, the spheroid cultures were optimized for a 384-well format in commercial HMM, a medium especially developed for hepatocyte cultures. Spheroid formation was consistent using seeding densities of 5,000–10,000 cells per well, whereas spheroids with seeding densities of 2,000 and below showed a more inconsistent morphology ([68]Figure 1A). A seeding density of 5,000 cells was chosen for all further studies to maximize oxygen and nutrient access to the interior of the spheroid. Figure 1. [69]Figure 1 [70]Open in a new tab Spheroid formation and basic hepatocyte functions (A) A, left: Different seeding densities after six days in culture in HMM. A, right: Spheroid formation in the eight media (seven days in culture, seeding density of 5,000 cells/well). (B) B, left: Average albumin secretion over 3 weeks of culture. Albumin production in vivo ranging from dormant to maximal albumin levels (gray zone) were calculated from literature values as outlined in the methods section. B, right: albumin secretion reported from different culture models for comparison (∗marks non-human cellular origin; [71]Table S2). (C) Average change in albumin secretion from week 1 to week 3. (D) Average ATP content over 3 weeks of culture. (E) Change in ATP content from week 1 to week 3. The span between the dotted lines in c and e indicates stable albumin and ATP levels (within 35%). b-e, Average data from four donors, error bars are showing standard deviation and media abbreviations can be found in [72]Table 1. Culture media Two commercial media with undisclosed content were investigated: Hepatocyte Maintenance Medium (HMM) and Cellartis Power Primary HEP Medium (PPM), as were two conventional media with well-defined contents: William's E medium (WE[HG]) and DMEM/F12 (DF[HG]; [73]Table 1 in [74]STAR Methods). Table 1. Media supplements HMM PPM WE[HG] WE[NG] WE[NG+] DF[HG] DF[NG] DF[NG+] Insulin, ng/mL 10,000 ≈6,000 10,000 0.58 0.58 10,000 0.58 0.58 Glucose, mg/L ≈2000[75]^a ≈970[76]^a 2000 990 990 3,151 990 990 Zinc (ZnCl[2]), μg/mL NS NS NS NS 1 NS NS 1 Transferrin, μg/mL 5.5 NS 5.5 5.5 5.5 5.5 5.5 5.5 Selenium, ng/mL 5 NS 5 5 5 5 5 5 Dexamethasone, μM 0.1 NS 0.1 0.1 0.1 0.1 0.1 0.1 Penicillin, U/mL 100 100 100 100 100 100 100 100 Streptomycin, μg/mL 100 100 100 100 100 100 100 100 L-glutamine, mM NS NS 2 2 2 2 2 2 [77]Open in a new tab NS, Not supplemented to the medium. ^a Measured values not supplied by the manufacturer. WE[HG] and DF[HG] (like most conventional media) are hyperglycemic with a glucose concentration of 2,000 mg/L and an insulin concentration of 10,000 ng/mL, both far above the normal fasting state (700–1,000 mg/L for glucose and 0.19–1 ng/mL for insulin ([78]American Diabetes Association, 2017; [79]Williams, 2016). Therefore, we modified these media by reducing the glucose and insulin to fasting levels and named these media variants William's E normoglycemic medium (WE[NG]) and DMEM/F12 normoglycemic medium (DF[NG]), respectively ([80]Table 1 in [81]STAR Methods). Because insulin is supplied in a concentrated zinc solution, the reduction of insulin solution to fasting levels also decreased the zinc concentration below physiological levels. Therefore, we modified WE[NG] and DF[NG] by adding back zinc ions to normal plasma levels (0.7–2.5 μg/mL) ([82]Maxfield and Crane, 2020) and named these media variants WE[NG+] and DF[NG+], respectively. Thus, eight different media were investigated ([83]Table 1 in [84]STAR Methods). Albumin secretion and ATP content The average albumin secretion was within the normal range observed in vivo ([85]Figure 1B) and comparable to, or greater than, that observed in other PHH models ([86]Davidson et al., 2016; [87]Hu et al., 2018; [88]Leite et al., 2016; [89]Ong et al., 2017; [90]Ramaiahgari et al., 2014; [91]Toba et al., 2020; [92]Török et al., 2011; [93]Tostões et al., 2012). Albumin secretion was stable (defined here as within +/− 35% of first week) over time for the two commercial and the three WE-media, but lower from the spheroids cultured in the DF media ([94]Figure 1C). No systematic differences in albumin secretion were observed between the hyper- and normo-glycemic WE and DF media. The commercial media had spheroids with the highest ATP content, followed by the WE media. The three DF media had spheroids with the lowest ATP content of all the media ([95]Figure 1E). For the two commercial media (HMM and PPM), the ATP content of the spheroids remained stable over the three weeks in culture ([96]Figure 1D), whereas it was reduced two- to three-fold in the other media. As for the albumin secretion, no systematic differences could be seen between the hyper- and normoglycemic media. Spheroid morphology and immunohistochemistry We then used immunohistochemistry to allow a more detailed morphological assessment of PHH spheroids cultured in the different media. In contrast to the light microscopy ([97]Figure 1A), the stained spheroid sections from DF media ([98]Figure 2A) shows asymmetric and fragile 3D structures compared to those in the other media. Thus, spheroids cultivated in the two commercial and WE-media consistently formed spheroids of the expected rounded shape ([99]Figure 2A). Figure 2. [100]Figure 2 [101]Open in a new tab Morphology and status of PHH spheroids cultured for three weeks in different media The center of the spheroids was selected for analysis. (A) Hematoxylin and eosin staining (H & E). (B) Adipolipin (PLIN2) staining for identification of lipid deposits. (C) Pimonidazole (Pimo) staining for assessment of hypoxia. (D) Caspase 3 staining for apoptosis. (E–G) Quantification of stained areas in e, PLIN2; f, Pimo; and g, Caspase 3 (Casp 3), respectively. Five spheroids were used for quantification of average and standard deviation in each medium and images are representative for the spheroids. Scale bar = 100 μm. In hepatocytes, hyperglycemic levels of glucose and insulin result in abnormal ectopic accumulation of lipids – a feature of steatosis ([102]Parks, 2002). We therefore stained for adipolipin (PLIN2), a protein located around the periphery of lipid droplets, to investigate the development of a steatotic phenotype ([103]Parry and Hodson, 2020). The PLIN2 staining confirmed the steatotic phenotype after cultivation in hyperglycemic media ([104]Figure 2B) and its absence in normo-glycemic media ([105]Figure 2E). Further, an increase of hollow spherical structures, typical of glycogen deposits ([106]Aluko et al., 2020), was found in the hyperglycemic cultures ([107]Figure 2A). We next investigated the access to oxygen in the spheroids, because an uneven oxygen distribution could result in a hypoxic inner core, as seen in other, usually larger spheroids ([108]Senkowski et al., 2015). For the hypoxia marker, we used the well-used pimonidazole staining ([109]Figure 2C; ([110]Masaki et al., 2016)). The staining was significantly higher in the PPM medium and was evenly distributed with no indication of a hypoxic core. Because hypoxia may lead to apoptosis and/or necrosis, we also stained for the apoptosis marker caspase 3 ([111]Figures 2D and 2G). All spheroids displayed low levels of caspase 3 with no large differences among the media ([112]Figures 2D and 2G). Differences in global protein expression Next, we investigated the effect of the different media on the global protein expression. Freshly thawed primary human hepatocytes (PHHs) from four donors were used as references for native PHH since the